Edit: since some users raised the concern that the question cannot be answered with peer-reviewed articles, I looked for studies in PubMed, an exclusively academic database. The first two studies have been retrieved there.
- The hard data
Apparently, there have been scientific studies on the topic, although not many. The following two articles looked the most relevant to me.
This paper was published on the peer-reviewed journal "Advances in Health Sciences Education". It studies the effects of sandwich feedback on medical students on clinical patient note-writing assignments. From the Discussion:
These studies indicate that students think feedback sandwiches positively impact their subsequent performance when they do not. [...] We find a consistent placebo dose effect on student perceptions in that full sandwiches are perceived as most effective, followed by ‘open-faced’ sandwiches, followed by unsandwiched feedback. While substantive positive comments resulted in an interim improvement in congruence with calibration content scores this did not result in improved patient note scores in T2.
Does the disconnect between learner perception and impact on performance matter? That is, is there any negative consequence to students misperceiving the impact of sandwiches? Study 2 hints that feedback sandwiches containing more substantive positive comments might actually be detrimental to students’ ability to critically self-assess, since there was a positive relationship between the number of substantive positive comments at T1 and less accurate self-assessment at T2.
- This other article published in "Resuscitation", a journal of the Elsevier group, compares the learning conversation and the sandwich technique in a BLS (Basic Life Support) training course. While they don't differ much in their final effect, the paragraph about instructors feedback is interesting:
Nineteen (40.4%) instructors remarked that the sandwich technique was too structured and repetitive; “Very repetitive—students quickly picked up on pattern of technique and therefore began to ignore it and so it lost its value”. Sixteen (34%) instructors commented that the sandwich technique was awkward to use as candidates naturally wanted to talk about points for improvement first; “Students were quick to think of the negatives rather than the positives of their performance”.
- A more theoretical approach
This article ("The sandwich feedback method: Not very tasty") is a review on the topic: it provides a good insight on why sandwich feedback arose in the first place and why it isn't such a good idea. It has been published by the Journal of Behavioral Studies in Business, a peer-reviewed journal.
The study focuses on the sandwich feedback in the workplace, where it can become a repeated behaviour that involves the same people (manager and employees), but I think its conclusions can be extended to its use in common environments.
Why the sandwich feedback is used
[L]eaders admitted that they used this particular approach since
they find giving negative performance feedback too stressful. Managers found it to be more relaxing by beginning the discussion with the employee by starting first with positive comments.
[...]
Other reasons why leaders may use the sandwich method surround the issues of optimism and being positive. Managers are encouraged to be upbeat based on two fundamental motivational perspectives:
The approach perspective holds that individuals move toward those things they find
attractive. Matlin & Stang (1978) stated that there could possibly be no virtue more enviable in the United States than to be an optimistic and positive person.
[The article appears to give for granted that the sandwich approach was born in the USA.]
The avoidance perspective holds that individuals try to evade that which they find to be undesirable or disagreeable. This is the case with providing subordinates negative feedback. Such feedback, however, presents a dilemma; most believe it necessary but few want to deliver it
(Ilgen & Davis, 2000).
Why the authors consider the sandwich feedback to be ineffective
The article gives a thorough list of both speculation- and literature-based (though it's literature written by managers relying on their personal experience) of why they consider this method to be ineffective. In short, it benefits the manager and not the employee. Over time, moreover, the employee learns to anticipate a reproach when praised and will doubt the honesty of the praise itself - also because the positive qualities of a person will likely stay more or less the same, thus leading the manager to repeat themselves.
The article also proposes an alternative to the sandwich feedback. While it is obviously focused on a relationship involving hierarchies such as the one between a manager and their employees, I think it's worth reading it:
1.
Plan the discussion, when possible.
2.
Keep positives and negatives separate.
3.
Time discipline so as not to be too soon or too late.
4.
Focus on the issue regarding employee behavior.
5.
Connect the behavioral
issue to
how the issue
impacts
the business.
6.
State consequences if behavior does not improve.
7.
Identify the proper and required behavioral change that
the supervisor expects.
8.
Ask how
the
manager can help the worker.
9.
Express
confidence in the employee’s ability to improve.